Jeanette Spear and the Estate of her late husband, Leon Spear, ("taxpayers") appeal the decision of the United States Tax Court assessing substantial income tax deficiencies and fraud penalties against them following a five-day trial. The tax court's decision depends in significant measure on "deemed" facts resulting from a sanction imposed because of Jeanette Spear's failure to appear and testify at trial. These deemed facts were critical to the outcome because they not only furnished the predicate for use by the Internal Revenue Service ("IRS") of the net worth method to determine income tax liability, but also appear to have shifted the burden of proof on both net worth and fraud from the IRS to the taxpayer.

The tax court imposed this quite severe sanction notwithstanding that it had before it a five-hour long videotaped deposition of Jeanette Spear taken for possible use at trial which covered all the ground of reasonably expected trial testimony. Moreover, the ultimate basis for imposition of the sanction, Jeanette Spear's putative bad faith in failing to appear at trial, is based on such a frail foundation that the tax court's bad faith finding does not survive even deferential review. Given these considerations, and the fact that the other factors that we consider in applying the principles used to assess the validity of sanctions favor the taxpayers, we conclude that the sanction imposed here was improper and an abuse of discretion. We will therefore vacate the tax court's decision and remand for further proceedings.

During the years in question, taxpayers were the sole shareholders in several corporations which operated a large number of parking lots in Center City, Philadelphia on the fringe of the downtown area. The IRS contends that taxpayers skimmed money from these cash businesses and failed to report it as income. The IRS based its assessment of deficiencies on the net worth method, under which it determined income by subtracting taxpayers' net worth at the end of the tax year from their net worth at the beginning of the tax year with appropriate adjustment for nontaxable receipts and nondeductible expenditures. The IRS often uses this method when the taxpayers' income and expense records are inadequate or incomplete.

In 1986, the IRS issued a Notice of Deficiency to taxpayers assessing income tax deficiencies of $51,271.70, $157,706.46 and $93,536.23 for the years 1975, 1976 and 1977 respectively. The Notice also asserted fraud penalties of $25,635.85, $78,853.23 and $46,768.12 for the same years. JA 27-33. Taxpayers sought a redetermination of these assessments in tax court. On October 31, 1989, Leon Spear suffered a stroke and died soon thereafter. The tax court substituted the Estate of Leon Spear as a defendant.

The taxpayers contended at trial that: 1) the IRS had inappropriately used the net worth method because they had kept adequate records of their income; 2) the source of the funds that led to the large increase in their net worth was $380,000 in cash that Leon Spear's father had given to him years earlier which had been kept in safe deposit boxes, so that taxpayers' net worth at the beginning of the 1975 was far higher than the IRS believed; and 3) the parking lots could not have produced sufficient income to account for the increase in net worth the IRS claimed. The tax court rejected these contentions and concluded that there were tax deficiencies of $43,354.65, $155,504.29 and $92,053.20 for 1975, 1976 and 1977. It also imposed fraud penalties of $21,677.32, $77,752.14 and $46,026.60 for the same years.

Although taxpayers repeat on appeal their contentions about the use of the net worth method, and challenge the factual findings pertaining to net worth as clearly erroneous, they also strenuously argue that the court committed reversible error by sanctioning them for Jeanette Spear's failure to testify. The sanction was a linchpin of the tax court's decision, and we limit our discussion of the record to the facts bearing on the sanctions issue.

B. The Facts Leading to the Imposition of Sanctions

In April 1990, the tax court entered an order setting the case for trial on November 9, 1990. JA 5. The IRS subpoenaed Jeanette to appear at trial because she was the only living witness to the alleged 1957 gift of $380,000 from Leon's father, and because she had been responsible for maintaining the books of the parking corporations. JA 923-24.

On October 25, 1990, taxpayers moved for a continuance on the basis that Jeanette was experiencing emotional trauma based on the anniversary of her husband's death (a death she attributed to the prosecution by the IRS, JA 11-12) and the approach of the trial. On November 2, 1990, Dr. Sol B. Barenbaum, a psychologist chosen by the Commissioner, examined Jeanette and reported that she could testify without mental or physical harm. JA 10-13, 124-25. However, on November 5, 1990, Jeanette was admitted to the psychiatric unit of Nazareth Hospital in Philadelphia after her attending physician, Dr. Martin J. Durkin, was told that she had attempted suicide by gas and possibly pills. JA 16, 125.

Taxpayers then moved for a continuance, attaching a letter from Dr. Durkin, who is a Board-certified psychiatrist, stating that Jeanette was suffering from "psychotic depression and a recent serious suicide attempt" and that she needed to be hospitalized for at least two or three weeks. JA 14. The tax court granted the continuance on November 6. JA 7. The next day Jeanette's son, Robert Spear, requested that she be released from the hospital. The hospital allowed her out for the day on November 9, 10, and 11, and discharged her on November 12. JA 143-44.

Dr. Durkin evaluated Jeanette again in December 1990, and January, March and April, 1991. (JA 15-18). On March 18, Dr. Durkin wrote to defense counsel that after three psychiatric evaluations of Jeanette he had concluded that

[s]he continues to suffer from a depressive illness with features of anxiety. I do not feel it wise to expose the patient to a judicial process in respect to her concerns with the Federal Government. This type of exposure could exacerbate her present illness and possibly lead to another suicide attempt. The stress could be a precipitating event to a possible heart attack or stroke.

J.A. 15. On April 29, after conducting still another psychiatric evaluation, Dr. Durkin again wrote to defense counsel. He stated that based on his continuing personal evaluation of Jeanette combined with the evaluation of a neurologist and a recreational therapist who observed Jeanette and conducted several tests during her hospitalization, his

opinion remains that the patient should not be exposed to depositions or to interrogatories because of her present gradual emotional status. To again summarize, I believe that any type of exposure to these types of events would exacerbate her depression and again cause a psychiatric hospitalization. Worse, the patient may again make an attempt at suicide which could be successful.

JA 18.

There is no evidence that, after this April evaluation, Jeanette had any further treatment until the next time the case was set for trial. See T.C.Memo.Op. at 23. In July, 1991, the Commissioner sought leave to take a videotaped deposition of Jeanette, arguing that such a deposition was needed due to Jeanette's possible unavailability for trial as a result of "her mental, emotional or physical infirmity." JA 105. Taxpayers opposed the application, submitting the April letter from Dr. Durkin quoted above, in which Dr. Durkin noted Jeanette's obsession with the trial and that she suffered from transitional stress due to her difficulties with the IRS. JA 18.

On August 8, 1991, the IRS moved for a court-ordered physical and mental examination to determine Jeanette's ability to testify. Taxpayers opposed the application, submitting another letter from Dr. Durkin stating that a forced examination or appearance in a court would "be a serious danger to Mrs. Spear," JA 21, and pointing out that Jeanette had previously undergone a court ordered examination. Taxpayers also submitted an affidavit from Dr. Marvin Rubin, a psychologist who had treated Jeanette from November 15, 1989 through October 24, 1990, stating that:

[w]hether correctly or incorrectly Jeanette attributes the death of her husband to the fear and anxiety that he had relating to the Internal Revenue Service hearing. Jeanette is very anxious and upset about the possibility of herself dying at the hearing. It is my professional opinion that Jeanette is incapable presently to withstand the trauma of a court hearing due to her emotional and psychological state. Add in the fact that the anniversary of her husband's death is imminent, the effect would psychologically devastating.

J.A. 11-12.

The court denied the motion for a videotaped deposition but granted the motion for a physical and mental examination. Taxpayers refused to have Jeanette appear for the examination and the Commissioner moved for sanctions. See T.C.Mem.Op at 24. Taxpayers then decided that it was preferable for Jeanette to appear for the videotaped deposition than the physical examination. JA 165. On November 13, 1991, the court ordered Jeanette to appear for the videotaped deposition and scheduled briefing on the sanctions motion.

On December 12, 1991, the IRS deposed Jeanette (on videotape) for more than five hours. In taxpayers' submission, they gave the IRS great leeway in questioning, objecting only eight times and not asking any follow up questions in order not to elevate Jeanette's level of stress. The taxpayers contend that Jeanette was distressed and confused at times during the deposition; the IRS asserts that she showed that she could testify coherently and knowledgeably. Compare JA 650, 764-65, 807, 814 with JA 638-40, 659-63, 736-38, 750-52. At the conclusion of the deposition, IRS counsel asked that the transcript be marked for use at trial. JA 921. After the deposition, the court granted the IRS' motion to withdraw its request for sanctions and denied the IRS' motion for a competency hearing as moot. JA 185.

The IRS subpoenaed Jeanette to appear as a witness at trial. On February 17, 1992, taxpayers notified IRS counsel that Jeanette would not appear at trial because doing so would endanger her health and because the IRS had taken her videotaped deposition two months earlier. The Commissioner moved to compel Jeanette to testify or for sanctions. On February 20, the tax court held a hearing at which the Commissioner's counsel offered to limit Jeanette's testimony to one or two hours and to "hold it in an atmosphere similar to that of a deposition." JA 8, 835-36, 921, 926, 929. Taxpayers did not accept this arrangement, and the court ordered Jeanette to appear and testify on February 24. See T.C.Mem.Op. at 25.

Taxpayers requested that the court schedule an evidentiary hearing on February 25 for Dr. Durkin to testify about Jeanette's condition and thus to help the court understand why she could not testify. JA 837, 920. The court denied the request on the grounds that it would disrupt the trial schedule, see T.C.Mem.Op at 24, although counsel offered to have Dr. Durkin testify before Jeanette's scheduled testimony so as not to disrupt the trial. JA 930. The court agreed to accept another affidavit from Dr. Durkin instead, stating "[i]f it's just a matter of effectiveness of presentation, oral versus writing, then I'm not going to have a hearing for that purpose." JA 932.

On February 23, the day before she was to testify, Jeanette was again admitted to Nazareth Hospital. According to the hospital records, the accuracy of which Dr. Durkin certified as the attending physician, Jeanette had a "major depressive affective disorder, recurrent episode, severe with psychotic behavior" and "unspecified acute reaction to stress." JA 225. Dr. Durkin's admission note states that Jeanette cried frequently during the evaluation, had a hopeless demeanor, and had difficulty concentrating. JA 232-33. Although her sons reported that she may have been abusing valium, laboratory tests did not show the presence of valium or any similar substances in her system. JA 217, 232, 234-40, opinion at 26. Jeanette did not appear in court on February 24 and the tax court granted the Commissioner's motion for sanctions. JA 186, 1095.

On February 25, the day after she was supposed to testify, while the trial was still going on, Jeanette checked out of the hospital. The taxpayers did not inform the court that Jeanette had done so (T.C.Memo.Op. at 26).

As a result of Jeanette's failure to testify, the tax court sanctioned the taxpayers by deeming the Commissioner to have "made a prima facie showing" of the allegations in paragraph 7 of the Commissioner's answer, those dealing with net worth and fraud, JA 186, 1095-96 opinion at 33, and to have "met the burden of going forward as to those allegations. This shifts the burden of going forward with evidence to petitioners to rebut respondent's allegations of fraud." T.C.Mem.Op. at 33.

Among the facts the court deemed to be true were that: (1) taxpayers had furnished only incomplete tax records to the IRS; (2) the IRS had determined correct taxable income for the years 1975-77 on the basis of the net worth method; (3) taxpayers did not have available any cash on hand as of December 31, 1974 which was not deposited in one of their bank accounts; (4) taxpayers used unreported income to acquire eight real estate properties in their names or the name of a wholly owned nominee corporation used to conceal their real estate holdings, and also used unreported income for other expenditures; and (5) taxpayers understated their taxable income for the years 1975-77 with fraudulent intent.

In a net worth case, the Commissioner must: (1) establish with reasonable certainty an opening net worth; and (2) either (a) show a likely income source, or (b) negate possible nontaxable income sources. Holland v. United States, 348 U.S. 121, 132-38, 75 S. Ct. 127, 133-37, 99 L. Ed. 150 (1954), T.C.Mem.Op. at 36. By deeming the IRS to have established correct taxable income on the basis of the net worth method, the tax court appears to have shifted the burden of proof on this central aspect of the case. T.C.Mem.Op. at 33 (holding for the IRS generally, the court stated that "[p]etitioners did not provide sufficient evidence to overcome these deemed facts").

That the imposition of sanctions may well have been critically important to the result is especially clear when considering the fraud counts. The tax court did state that:

Respondent has the burden of proving fraud by clear and convincing evidence. Sec. 7454(a); Rule 142(b). First, respondent must prove the existence of an underpayment. Parks v. Commissioner, 94 T.C. 654, 660 (1990). Respondent may not rely upon the taxpayer's failure to carry the burden of proof as to the underlying deficiency. Parks v. Commissioner, supra at 660-661; Petzholdt [Petzoldt ] v. Commissioner, 92 T.C. 661, 700 (1989); Estate of Beck v. Commissioner, 56 T.C. 297, 363 (1971). Second, respondent must show that the taxpayer intended to evade taxes by conduct intended to conceal, mislead, or otherwise prevent tax collection. Petzholdt [Petzoldt ] at 699. Stoltzfus v. United States, 398 F.2d 1002, 1005 (3d Cir.1968); Parks v. Commissioner, supra at 661; Rowlee v. Commissioner, 80 T.C. 1111, 1123 (1983).

T.C.Memo.Op. at 55. Yet the tax court seemed to enable the Commissioner to surmount this steep burden of proof by relying on the facts deemed to be true. The court wrote:

On February 24, 1992, the Court imposed sanctions on petitioners because Jeanette Spear violated an order of the Court by unreasonably refusing to testify at trial. The Court ordered that respondent is deemed to have made a prima facie showing that the facts in paragraph seven of the amended answer (paragraph 7) are established. Petitioners did not convince us that any of the statements of facts in paragraph 7 are wrong. As discussed below, we conclude the facts stated in paragraph 7 and the entire record in this case clearly and convincingly show that Leon and Jeanette Spear are liable for fraud for each year in issue.

T.C.Memo.Op. at 55-56.

Despite these indications that the tax court switched the burden of proof as well as the burden of production, there are many other places in the opinion that make it appear that the tax court found sufficient evidence of net worth and of fraud without relying on the deemed facts. Nonetheless, because we are unsure whether the court relied on these facts and shifted the burden of proof, and because the consequences to the taxpayers are so significant, we must assume that the court did rely on these facts. We will thus treat the sanction as one that essentially shifted the burden of proof (and production) on net worth and on fraud.

We note that shifting the burden of proof on the fraud counts would be an even more severe sanction here than it would be ordinarily because the tax court relied on taxpayers' fraud to reject their statute of limitations defense. See T.C.Mem.Op. at 62. Fraud will defeat a statute of limitations defense, Sec. 6501(c)(1), and if taxpayers had prevailed on the fraud count, they may well have had a valid statute of limitations defense.1 See T.C.Mem.Op. at 62-63. In sum, the sanction in this case was quite significant and may well have controlled the outcome.

T.C. Rule 104(c) as quoted in Gerling Intern. Ins. Co. v. C.I.R., 839 F.2d 131, 136 n. 7 (3d Cir.1988), provides in pertinent part that:

"If a party ... fails to obey an order made by the court ... the Court may make such orders as to the failure as are just and among others the following:

(1) An order that the matters regarding which the order was made or any other designated facts shall be taken to be established for the purposes of the case in accordance with the claim of the party obtaining the order.

The rule is quite similar to Fed.R.Civ.P. 37(b)(2), and we construe them in pari materia. We review the sanction of deeming facts to be true under an abuse of discretion standard. See Ins. Corp. of Ireland v. Compagnie Des Bauxites, 456 U.S. 694, 707, 102 S. Ct. 2099, 2107, 72 L. Ed. 2d 492 (1982); Ali v. Sims, 788 F.2d 954, 957 (3d Cir.1986).

In the context of discovery abuse, the Supreme Court has provided guidance on use of the sanction of deeming facts to be established. In Ins. Corp. of Ireland, 456 U.S. at 707, 102 S.Ct. at 2107, the Court explained that

Rule 37(b)(2) contains two standards--one general and one specific--that limit a district court's discretion. First, any sanction must be 'just'; second, the sanction must be specifically related to the particular 'claim' which was at issue in the order to provide discovery.

In that case the Court held that the district court had not abused its discretion in deeming facts establishing personal jurisdiction to be true absent proof to the contrary, because defendants had repeatedly agreed to comply with discovery orders and then failed to do so despite warnings that sanctions would result. Ins. Corp. of Ireland at 707-09, 102 S. Ct. at 2107. The Court held that the second requirement, that the sanction be related to the claim at issue, was met because the sanctions took as established facts that plaintiff was seeking to prove through discovery.

This court has not elaborated on or applied the Insurance Corp. of Ireland standard. In Ali, supra, we held that where a district court sanctioned defendants by deeming allegations in plaintiff's complaint to be admitted and granted summary judgment for plaintiff, the ruling was equivalent to a default judgment and thus required application of the standards we had set for issuing a sanction of dismissal. See 788 F.2d at 957. More specifically, we held in Ali that, under the factors we had articulated in Poulis v. State Farm Fire and Casualty Co., 747 F.2d 863 (3d Cir.1984), the sanctions constituted an abuse of discretion. In Poulis we had explained that our review of a district court's dismissal with prejudice "is guided by the manner in which the trial court balanced [six] factors ... and whether the record supports its findings." Poulis, 747 F.2d at 868. The six factors are:

(1) the extent of the party's personal responsibility; (2) the prejudice to the adversary caused by the failure to meet scheduling orders and respond to discovery; (3) a history of dilatoriness; (4) whether the conduct of the party or the attorney was willful or in bad faith; (5) the effectiveness of sanctions other than dismissal which entails an analysis of alternative sanctions; and (6) the meritoriousness of the claim or defense.

Id.

In Ali we applied these factors to reverse a sanction deeming certain facts to be true. We held that, even if there was inexcusable delay by the defendants in that case, there was no bad faith, no history of dilatoriness, little prejudice from the delay that was caused, and less severe sanctions were probably available. Under those circumstances, sanctions that were equivalent to dismissal constituted an abuse of discretion. Id. at 957-58. We explained that, "[i]n Poulis, we established the strong presumption against sanctions that decide the issues of a case." Id. at 958.2

Here, unlike in Ali, the tax court's sanction did not end the case. At most the tax court deemed certain key facts admitted and reversed the burden of proof. While this is a severe sanction, it is not the same as deeming allegations in a complaint to be admitted or granting a default judgment. In Chilcutt v. United States, 4 F.3d 1313 (5th Cir.1993), the Fifth Circuit considered the standards for imposing a similar sanction (of deeming prima facie elements of the plaintiffs' liability claim to be established). The court held that, although a court's decision to deem certain facts established may sometimes be equivalent to a default judgment, it was not equivalent where the sanctioned party (the government) was allowed to present its case in chief and could have prevailed if it had established its contentions by a preponderance of the evidence. Id. at 1320 & n. 18. Thus, instead of imposing the sanction under the standards appropriate for a dismissal, the court applied the two standards of Insurance Corp. of Ireland (the requirement of "justness" and the requirement that the sanction be related to the particular claim at issue in the order to provide discovery)--"along with a third--that the sanction meet the Rule 37 goals of punishing the party which has obstructed discovery and deterring others who would otherwise be inclined to pursue similar behavior." Id. at 1321.

Because the sanction was not equivalent to default, for which a prerequisite under Fifth Circuit law is flagrant and willful disregard, the court suggested, in what it admitted to be dicta, that flagrant and willful disregard was not necessary. Id. at 1322 n. 23. On the facts of the case, the Chilcutt court upheld the sanction. It stated that, where the district court had warned the government that it would issue sanctions and the government had repeatedly promised to be forthcoming, the plaintiffs had a colorable claim, and the evidence the government had hidden was relevant to the plaintiff's case, the sanction was just, related to the claim sought to be proved, and was necessary to compensate for non-compliance and to deter future violations. As for other considerations, the government's conduct was willful and was not solely the fault of its attorney. Id. at 1321-25.

We agree with the Chilcutt court that cases on the sanction of dismissal are not automatically applicable to the sanction of deeming certain facts to be established. Nonetheless, the Poulis factors are relevant to evaluating such a sanction. This is clear from the fact that, in evaluating whether a district court has properly exercised its discretion in imposing the sanction of exclusion of testimony, a sanction less harsh than dismissal and probably similar to shifting the burden of proof, we consider factors similar to those in Poulis. See Meyers v. Pennypack Woods Home Ownership Ass'n, 559 F.2d 894 (3d Cir.1977). We consider:

(1) the prejudice or surprise in fact of the party against whom the excluded witnesses would have testified, (2) the ability of that party to cure the prejudice, (3) the extent to which waiver of the rule against calling unlisted witnesses would disrupt the orderly and efficient trial of the case or of other cases in the court, and (4) bad faith or willfulness in failing to comply with the district court's order.

Id. Meyers and Poulis supply the sources of the standard we adopt here.

Comparing the Meyers factors with Poulis, Factor 4 goes to a party's culpability as do factors 1, 3, and 4 of the Poulis factors. See supra at 12. Factors 1 and 2 go to prejudice as does factor 2 of Poulis. Factor 2 also goes to the ability to correct the problem with action less harsh than the sanction being considered as does factor 5 in Poulis. Moreover, just as the ultimate Poulis calculus is a balancing, we think that balancing similar factors is appropriate in assessing a sanction of deeming established certain facts. We apply a sliding scale--the harsher the sanction being imposed, the more the balance will have to be against the party being sanctioned to justify the sanction. See National Hockey League v. Metropolitan Hockey Club, Inc., 427 U.S. 639, 643, 96 S. Ct. 2778, 2781, 49 L. Ed. 2d 747 (1976) (dismissed); Societe Internationale Pour Parcipations Industrielles et Commerciales v. Rogers, 357 U.S. 197, 212, 78 S. Ct. 1087, 1096, 2 L. Ed. 2d 1255 (1958) (dismissal); Donnelly v. Johns-Manville Sales Corp., 677 F.2d 339, 342-43 (3d Cir.1982) (dismissal); Meyers v. Pennypack Woods Home Ownership Ass'n, 559 F.2d 894 (3d Cir.1977) (exclusion of critical evidence).

This approach is consistent with the Fifth Circuit's opinion in Chilcutt. Although the Chilcutt court held that the dismissal cases were not on point and that the test from Insurance Corp. of Ireland applied, the court referenced all of the factors we consider in dismissal cases. It considered the culpability of the sanctioned party including whether the violation was solely the fault of the attorney or was also the fault of the client, and the effectiveness of alternative sanctions. And while the court stated that willfulness was not required to impose a sanction of deeming facts proved (thus, incorporating our sliding scale theory of the appropriateness of sanctions), it also implied that willfulness was relevant. It stated that "of course, the flagrancy of a party's behavior must be directly proportionate to the severity of the sanction imposed." Chilcutt at 1322 n. 23.3

This approach is also consistent with Insurance Corp. of Ireland itself. The standard articulated there, that a sanction must be 1) just and 2) specifically related to the particular 'claim' which was at issue, was essentially a general standard for all Rule 37 sanctions. Thus, like our opinion here, our opinions in Poulis and Meyers had to be consistent with Insurance Corp. of Ireland because they involved Rule 37 sanctions. They were consistent with it because they were an elaboration of the meaning of "just" and "related to the particular claim" in particular contexts.

In sum, in reviewing a trial court order deeming evidence admitted as a sanction for litigation misconduct, we will engage in a weighing and balancing exercise in which we consider: 1) culpability (including willfulness and bad faith, and whether the client was responsible or solely the attorney); 2) prejudice; and 3) whether lesser sanctions would have been effective. In making the actual balancing we utilize a sliding scale, so that bad faith, for example will have to be quite high to tip the balance if other factors strongly favor the taxpayers. We view this exercise to be a transliteration of the Insurance Corp. of Ireland standard in that the prejudice consideration subsumes the specific relatedness requirement, all of the factors of which essentially elaborate on "justness."

1) The Need to Show Bad Faith or Willfulness.

In the jurisprudence of dismissal, willfulness or bad faith is almost always required in order for dismissal to be within the proper scope of the court's discretion. In the particular cases before it, the Supreme Court, at a minimum, has required some sort of fault for dismissal to be allowable. Compare Societe Internationale Pour Parcipations Industrielles et Commerciales v. Rogers, 357 U.S. 197, 212, 78 S. Ct. 1087, 1096, 2 L. Ed. 2d 1255 (1958) (where party could not comply with discovery order due to Swiss law, dismissal was inappropriate. It was "due to inability and not to willfulness, bad faith, or any fault of petitioner.") with National Hockey League v. Metropolitan Hockey Club, Inc., 427 U.S. 639, 643, 96 S. Ct. 2778, 2781, 49 L. Ed. 2d 747 (1976) ("[D]ismissal was appropriate in this case by reason of respondents' 'flagrant bad faith' and their counsel's 'flagrant disregard' of their responsibilities.").

Some courts have held that willfulness or bad faith is always required before dismissal is an acceptable sanction. See Ford v. Fogarty Van Lines, Inc., 780 F.2d 1582, 1583 (11th Cir.1986) ("Absent a clear record of delay or contumacious conduct by the plaintiff, the trial court's discretion is limited to the application of lesser sanctions [than dismissal]."); Wilson v. Volkswagen of America, Inc., 561 F.2d 494 (4th Cir.1977); Telectron, Inc. v. Overhead Door Corp., 116 F.R.D. 107, 131 (S.D.Fla.1987). But see United States v. Sumitomo Marine & Fire Ins., Co., 617 F.2d 1365, 1369 (9th Cir.1980) (although government did not exhibit bad faith, dismissal was necessary to deter flagrant disobedience that resulted from understaffing).

Although we have held that dismissals are an extreme sanction reserved for cases comparable to National Hockey League where there was flagrant bad faith, see Poulis, 747 F.2d 863, 867-68, we have sometimes upheld a court's sanction of dismissal even when there was no willfulness or bad faith. See Poulis, 747 F.2d at 868-70; cf. Hicks v. Feeney, 850 F.2d 152, 156 (3d Cir.1988) (not all Poulis factors have to be present for dismissal). Nonetheless, we generally have not upheld dismissal absent willfulness and bad faith. See Donnelly v. Johns-Manville Sales Corp., 677 F.2d 339, 342-43 (3d Cir.1982) (dismissal was an abuse of discretion where there was delay in obtaining local counsel but it was due to failure to move with dispatch rather than to bad faith, where the delay caused little prejudice to the defendant, and where the district court did not consider an alternative sanction). Even with respect to the less extreme sanction of exclusion of evidence, we have held that with respect to critical evidence, "the exclusion of critical evidence is an 'extreme' sanction, not normally to be imposed absent a showing of willful deception or 'flagrant disregard' of a court order by the proponent of the evidence." Meyers at 905 (quoting Dudley v. South Jersey Metal, Inc., 555 F.2d 96, 99 (3d Cir.1977)).

Although, like the Chilcutt court, we do not have to decide the issue, we assume that, when the sanction of deeming facts to be true is not the equivalent of dismissal, willfulness and bad faith are not prerequisites for imposing that sanction. When a party does not provide information to another party to which that party is entitled, a court is certainly permitted to "even out" the proceedings by shifting the burden of proof in a fair way even in the absence of bad faith. Moreover, in Insurance Corp. of Ireland, the Supreme Court upheld a sanction of deeming facts to be established, even though the court had made no explicit finding of bad faith, finding repeated violations of discovery orders to constitute sufficient fault to justify the sanction.

Nonetheless, the presence of willfulness and bad faith certainly enhances the case for sanctions. Shifting the burden of proof, as the tax court seems to have done here when it deemed certain facts to be established, is a fairly extreme sanction. It significantly changes the likely outcome at trial. In the absence of willfulness or bad faith, other factors would have to weigh strongly in the favor of such a severe sanction to justify it.

2) Did Jeanette's Conduct Constitute Bad Faith?

The tax court's decision to sanction taxpayers was essentially bottomed on its finding that Jeanette was deliberately attempting to avoid her testimonial duties. The finding of bad faith was grounded upon the following: 1) Jeanette's illness and hospitalizations correlated with the time of her scheduled testimony; 2) the lack of evidence of illness at other times; 3) the taxpayers' agreement to a deposition after the court ordered a mental and physical examination of Jeanette, while initially maintaining that Jeanette could not be deposed; 4) Jeanette's competent answers to questions at her deposition, demonstrating a significant knowledge of the case; 5) taxpayers failure to timely inform the court that Jeanette could not testify (twice) or see a psychiatrist (once); and 6) taxpayers failure to inform the court that Jeanette had been released from the hospital until after the trial was over. T.C.Mem.Op. at 27-29.

A different explanation exists, however, for the correlation between the severity of Jeanette's illness and the imminence of court appearances (and the resultant tardiness of taxpayers informing the court that Jeanette could not testify) from the one that posits that the alleged illness was a tactic to avoid testifying. A legitimate, medically grounded connection may have existed between Jeanette's illness and the imminence of court appearances. In order to disbelieve this explanation (and believe that the correlation demonstrated that Jeanette was making up the illness to avoid court appearances), the tax court had to entirely discredit the evaluation of several physicians.

Dr. Durkin, a board-certified psychiatrist and apparently neutral witness who had never treated Jeanette prior to her admittance to the emergency room at Nazareth Hospital, consistently diagnosed Jeanette as having a major depressive disorder, and did so upon Jeanette's admission to the hospital shortly before trial. He also maintained that this disorder was related to her difficulties with the IRS, and he stated on several occasions between November, 1991 and August, 1992, that testifying would pose a serious threat to Jeanette's health. JA 15-JA 18. Dr. Durkin reached this conclusion based on several examinations of Jeanette, including examinations during a time when trial was not imminent. Moreover, he based his opinion not only on his own evaluation but on that of a neurologist and a recreational therapist who had examined Jeanette during her first hospitalization.

Dr. Durkin's evaluation was corroborated by the affidavit of Dr. Rubin, who also concluded that Jeanette's illness was related to the legal proceedings. JA 11. He stated that she attributed her husband's death to those proceedings and feared dying herself as a result of them. He added that testifying would be psychologically devastating to her. Like the tax court, these physicians were certainly aware of the possibility that Jeanette was feigning illness in order to avoid testifying, and yet they opined to the contrary. The only doctor who concluded that Jeanette was capable of testifying did so before her first hospitalization.4

The tax court's other justifications for its findings also do not demonstrate bad faith. The fact that taxpayers eventually agreed to allow Jeanette to be deposed does not show that their concern with her appearance at court proceedings was not genuine. Taxpayers were faced with a choice of having Jeanette submit to a mental and physical examination, having her deposed, or facing a significant possibility of sanctions. Their reluctant agreement to a deposition does not demonstrate a lack of concern that such a deposition would affect Jeanette's health.

We have viewed the videotape deposition which we describe infra at 115-16. Although Jeanette broke down and cried and had to be soothed on several occasions, and seemed confused as to questions at others, she basically gave a lucid deposition without emotional breakdown, a factor that, as the tax court noted, would seem to undermine the doctors' conclusions that testifying would be emotionally devastating to her.5 But we are not physicians. The deposition revealed Jeanette to be quite emotionally upset, and we cannot say with assurance that the fact that she was able to testify on one occasion automatically means that she could always do so.

Dr. Durkin continued to believe after this deposition (at the time of her second hospitalization) that Jeanette was suffering from a major depressive order. And he reported that upon hospitalization "she indicated that she became quite anxious and quite upset when she discovered that the Internal Revenue Service wished her to be deposed for another hour period of time. She add[ed], '[t]hey have all that they can possibly get from me, what else are they looking for.' " JA 232. Thus, Jeanette's fear of the IRS may have escalated after the videotaped deposition. Moreover, although the IRS offered to make the conditions during Jeanette's trial testimony similar to those during her deposition, the judge's presence at trial would have added an intimidating factor not present during the earlier deposition.

Finally, defendants' failure to inform the tax court of Jeanette's release from the hospital, while improper, does not show that taxpayers were deliberately creating an excuse to avoid having Jeanette testify. Jeanette was discharged against medical advice, JA 225, and if taxpayers had really been attempting to deceive the court, they would have had Jeanette stay in the hospital until the conclusion of the trial.

Thus, in the face of contrary opinions by two experts who had significant opportunity to examine Jeanette, the tax court's explanation for its finding that Jeanette's "refusal to testify was a manipulation, and not a bona fide response to medical problems," JA 29, was extremely thin. Moreover, it seems extremely unlikely that Jeanette was attempting to manipulate the trial process given that she had very little to gain by doing so--there is little reason to believe that her testimony would have substantially aided the Commissioner; rather, it might have significantly hurt the Commissioner. For example, based on our viewing of the deposition, we find Jeanette's testimony as to the $380,000 cash hoard quite straightforward, and it seems to be credible. Finally, we note that even if Jeanette initially went to the hospital partly as an attempt to avoid testifying and helping the Commissioner, after receiving medical advice, she had every reason to worry about testifying.

Similarly, Jeanette's co-defendants, the representative of the Estate of Leon Spear and ultimately of Jeanette's children, had every reason to worry about the effect that testifying would have on her. Thus, it is highly unlikely that taxpayers' refusal to allow Jeanette to testify was based solely on an attempt to manipulate the trial process and did not reflect significant concern for her health. Compounding the problem is the fact that the tax court declined to hold a hearing on the issue because such a hearing would allegedly have disrupted the trial schedule. Yet taxpayers offered to produce Dr. Durkin at the time when this would not disrupt the trial. Especially in the absence of a hearing at which the tax court could ask Dr. Durkin why he was sure that Jeanette was not making up her illness or at least its severity, the tax court's finding of bad faith is seriously problematic.

We acknowledge that taxpayer was not prevented from complying with the court's order due to an external constraint. Even if Jeanette truly feared becoming more sick if she testified, she still was physically capable of testifying and consciously chose not to do so. In this sense, her non-compliance was willful. Moreover, it was a choice that she made rather than a choice her attorney made. Nonetheless, we think that Jeanette's level of culpability was not high, given that we have found that her fears of testifying were legitimate.6

We review the tax court's finding of bad faith and wilfulness deferentially, i.e., for clear error. See Commissioner v. Duberstein, 363 U.S. 278, 290-91, 80 S. Ct. 1190, 1199-200, 4 L. Ed. 2d 1218 (1960); B.B. Rider Corp. v. Commissioner, 725 F.2d 945, 948 (3d Cir.1984); De Cavalcante v. Commissioner, 620 F.2d 23, 26 (3d Cir.1980). A finding is clearly erroneous when although there is evidence to support it, the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed. United States v. U.S. Gypsum Co., 333 U.S. 364, 395, 68 S. Ct. 525, 541, 92 L. Ed. 746 (1948). While we understand the tax court's annoyance with Mrs. Spear, for the reasons we have elaborated on at such length, we are left with such a firm conviction here. But even if the problematic bad faith finding survives because of deferential review (and if it did, it would be only by a small margin), the result would be the same because, under the sliding scale, the bad faith will have to be quite high to tip the balance in favor of the IRS in view of the fact that the other factors in the Insurance Corp. of Ireland-based test we apply strongly favor the taxpayer, see infra at 115-17, and it is not.

3) The Need to Show Prejudice.

In Insurance Corp. of Ireland, the Court held that "the sanction must be specifically related to the particular 'claim' which was at issue in the order to provide discovery." 456 U.S. at 707, 102 S. Ct. at 2107. It may be that this requirement does not inherently bar sanctions where there is no prejudice. Arguably a sanction may be considered to be "specifically related to the particular 'claim' " at issue in the discovery order even when there is little indication that the discovery would have produced useful information. A party should certainly not be able to gain a strategic advantage at trial by refusing to provide information it is required to provide and avoiding sanctions because the other side cannot demonstrate the importance of this information.

Nonetheless, the basic thrust of the Supreme Court jurisprudence is that sanctions that effect the outcome of the trial should only be imposed in order to compensate for violations that may plausibly be thought likely to affect the outcome of the trial. See Wilson v. Volkswagen of America, Inc., 561 F.2d 494 (4th Cir.1977) ("Even in those cases where it may be found that failure to produce results in the discovering party's case being jeopardized or prejudiced, it is the normal rule that the sanction must be no more severe than is necessary to prevent prejudice to the movant.") (quotations omitted).7 And so we conclude that the imposition of any sanction that affects the likely outcome of a trial requires that the party sanctioned have gained some advantage from his or her disobedience of a court order. In other words, the party that gains from the sanction must have been at least arguably prejudiced by the misconduct of the other side.

4) Was There Prejudice?

While the deposition was palpably an emotional experience for Jeanette, and she broke down and cried several times, her deposition was lucid and informative. As we viewed the deposition, she possessed and was able to and did relate, in response to questions, a great deal of information about the affairs of the parking lot business. There were also many things that she did not know or remember, but it seems unlikely that, given that the events happened so many years ago, she would recall additional details at trial. While Jeanette was quite deliberate, and sometimes stated that she did not understand what appeared to be simple questions (which were then repeated and answered), she was direct and composed. She was apparently not feeling well physically (as well as emotionally), but, given the length of the deposition and the amount of detail covered, it is difficult to see how anything more would be forthcoming at a trial.

In our view, there was no prejudice to the IRS from Jeanette's failure to testify. While the IRS stresses the need to obtain the truth, the fact is that the court received Mrs. Spear's 257-page (videotaped) deposition that we have described. What more did it need? To repeat, having viewed the videotape, we cannot conceive what more the IRS could have adduced at trial. Moreover, Jeanette's testimony was favorable to the taxpayers rather than the government; thus, the only thing the government had a reasonable chance of gaining from her testimony was a hope to trip her up a la Perry Mason and diminish the credibility of taxpayers' evidence. That rarely happens in the real world, and the Commissioner already had been presented with a chance to question Jeanette in a five hour deposition that occurred two months before trial during which defendants had very few objections to the questions posed by the Commissioner's counsel. The tax court thus had an excellent opportunity to judge Jeanette's credibility even without her appearance at trial.

Moreover, when the IRS requested the videotaped deposition, it did so in part because it was aware that Jeanette might not be available for trial and it marked the transcript for use at trial. Thus, during the deposition, the IRS had every incentive to ask all the questions it wanted to ask at trial. The IRS did not explain what additional questions it had for Jeanette that she had not already answered during the deposition.

The IRS argues that there was prejudice because Jeanette was a "key witness in this case." While Jeanette was a key witness, this does not explain why the IRS needed her live testimony. Although live testimony is generally preferable to videotaped testimony, the absence of such testimony, even from a key witness, is only minimally prejudicial when that witness is adverse and when there is a videotaped deposition that can be introduced in lieu of live testimony. That videotaped deposition testimony is a staple of modern case management in federal courts is too well established to require citation. And yet, as taxpayers contend, "[e]ssentially, the IRS claims that it was crucial to have Jeanette testify for a second time so that she would not be believed." (Appellant's Reply at 19). But that, we have noted, is no basis for a conclusion of prejudice.

5) The Balancing Exercise.

In view of the foregoing discussion, the balancing exercise is not difficult. We have concluded that the IRS incurred no prejudice, in view of the availability of the videotaped deposition. On the subject of whether lesser sanctions would have been effective, this does not seem to be a factor here. Although a finding of bad faith may not be strictly necessary to support sanctions, see Ins. Corp. of Ireland, 456 U.S. at 707, 102 S.Ct. at 2107; Hammond Packing Co. v. Arkansas, 212 U.S. 322, 350-51, 29 S. Ct. 370, 380, 53 L. Ed. 530 (1908); Meyers v. Pennypack Woods Home Ownership Ass'n, 559 F.2d 894, 905 (3d Cir.1977), the imposition of sanctions in the absence of bad faith generally requires a strong showing of prejudice. But whatever rationale the tax court judge might have had, there certainly were lesser sanctions than were employed here that could have "sent the message." Finally, we have concluded that the tax court's finding that Jeanette's failure to appear for trial was in bad faith is clearly erroneous, but that, even if not, it was sufficiently marginal that it would have been outweighed by the other factors which strongly militated in favor of the taxpayers' position. Hence, on the sliding scale the result is the same. Accordingly, the sanction constituted an abuse of discretion and must be set aside.

We have concluded that the sanction imposed by the tax court, of deeming admitted the facts that furnished the predicate for use of the net worth method, and shifting the burden of proof on both net worth and fraud from the IRS to the taxpayer, constituted an abuse of discretion and must be set aside. We will therefore vacate the decision of the tax court and remand the case for further proceedings consistent with this opinion. The court may, of course, elect to retry the case. In that event, it might be well advised to rely upon Jeanette's videotaped deposition in lieu of her testimony, although perhaps her emotional state is now better. On the other hand, the court may simply prefer to decide the case on the basis of the existing record, but absent the "deeming" and its consequences which we have declared invalid.8